I agree which is why I think remote work is the way to go for jobs that it can work for.
However, I think that guy was talking about industry pollution messing up the environment around. But I think that's a separate issue entirely and needs to be handled by another set of laws.
I think the problem here is there's no prior consent. If you've DM'd someone in the past, they can now call you. Twitter never had video call before, and the people who use it never expected to be giving consent for a video call maybe even several years later.
Current AI's ability to augment coding is no joke. I had ChatGPT (not even the latest one) write me a prosody plugin in Lua for some Jitsi integrations I was working on. It got it almost right, but I had to go in and compare its work with other prosody plugins to see where it went wrong. But see, at that point the only thing I knew about prosody was how to install plugins for it. Now I have a good understanding of how it works and also how it works with Jitsi based mostly on ChatGPT's code and explanations. This all happened over the course of maybe two hours whereas if I just did it by myself I'd probably be banging my head all day cause prosody's docs are passable but its documentation with regards to Jitsi's usage of it is non existent.
They definitely have a place in creative work. As an engine not a driver. For example, when writing songs I use them to help me try and rephrase things or find different words to convey what I want. I've also used them to give me ideas for variations to explore on game mechanics, or generate sample data with a loose set of parameters, or create some reference material for drawing. But every time I've used them they have pretty much never gotten it right the first time. Always there's revisions, always there's at least some massaging I have to do on my part to make it coherent. And why wouldn't there be? If I could write a prompt so precise that it gave me something exact, I would need to already know precisely what I want and maybe wouldn't even need the AI at all. For creative stuff at least, it has to make some garbage otherwise it's not really creating enough variation to be useful.
Pretty much the reason I never used it on my last phone. Even went through the steps to have it recognize my voice better, and even when it hears it the first time there's a 1 second delay before it actually starts listening, which makes me further contemplate why I bothered.
I'd still probably prefer the usual CLI for setup, commits, pushes etc. but this looks like a godsend for any branching/rebasing operations!